Author: emerson tan
Date: 23:59:20 03/14/04
Go up one level in this thread
On March 14, 2004 at 10:34:14, William Penn wrote: >On March 14, 2004 at 10:07:33, Michael Nolan wrote: > >>I apologize if this question has been asked many times before--I'm new to this >>forum. I am interested in doing computer-assisted analysis of games, and would >>like to get the strongest commercially-available software for time settings such >>as six or twelve hours per move. Possibilities are Shredder 8, Junior 8, Fritz >>8, Hiarcs 9, etc. Any suggestions? (In other words, I'm wondering if some >>programs are relatively much stronger at deep analysis than their 3 minutes / >>move ratings would indicate.) > >I have often wondered the same thing, but now believe it is impossible to answer >that question [my composition window glitch mentioned below just returned >here!?]. Comparing strengths of programs requires extensive tests to be reliable >[another composition window glitch here!?] such as SSDF, but it's practically >impossible to run such tests at long time controls like we want. So I believe we >have to settle for the test results from faster time controls [another shift of >composition window text to the left!?] such as SSDF uses, and keep our fingers >crossed. > We can test it thru a series of test positions that are deep and compare the results of the different engines to the correct answer. It should have plenty of positions representing tactical, positional and endgame. a thousand postions to test would require volunteers just like the SSDF because each position would be run for 12 hours. But first, we have to collect very deep positions with known correct solutions to test:) >At such long time controls in infinite analysis mode another special factor >becomes important [another composition window shift!?] too, related to how often >the analysis is displayed in the engine window. Do we have to wait 1 hour, 2 >hours, 10 hours, or exactly how long for the next "leg" of analysis to finish >then deign to display the result for user to see? I find that certain engines >are better in that regard, while others can take very long times between their >engine window outputs. Shredder seems to be one of the best in that regard. It >may [another composition window text shift to the left!?] also help to reduce >hash size, which seems to make those output intervals shorter. Of course that >runs counter to common scuttlebutt that longer time controls benefit most from >larger hash sizes. >WP
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.